Coding as We Know It May Be Over, Claims Anthropic’s Boris Cherny

Anthropic's Boris Cherny asserts that the traditional role of software engineers is rapidly becoming obsolete, with his company relying entirely on AI-generated code since last November. At the recent Code with Claude conference in San Francisco, he expressed frustration with the term "vibe coding," which many now use to describe this new AI-assisted programming approach. Cherny, who leads the development of Claude Code, believes this terminology fails to capture the significance of the transformation underway in the software engineering field.

The term “vibe coding” was popularized by OpenAI co-founder Andrej Karpathy in early 2025, quickly gaining traction and even being named the word of the year by Collins Online Dictionary in November. Cherny argues that calling AI-driven code generation vibe coding trivializes the substantial work accomplished by tools like Claude Code and OpenAI’s Codex, which produce millions of lines of production-grade code and generate billions in revenue. He likened it to calling surgery “gut feeling medicine,” stressing the need for a term that reflects the sophistication of the process.

Cherny has proposed the alternative term “agentic engineering,” another suggestion from Karpathy, but remains unsatisfied and has invited suggestions from the community. He is actively looking for a new label that accurately represents the shift from manual coding to AI-driven programming, welcoming ideas from attendees and readers alike.

The dialogue around terminology reflects a broader change within the tech industry. Cherny has stated that coding is effectively “solved” for him, as he manages to ship between 22 and 27 pull requests daily, all generated through AI. At Anthropic, Claude instances are designed to operate autonomously, communicating via Slack and executing engineering tasks with minimal human intervention. This level of automation is reshaping the responsibilities of software engineers, prompting a re-examination of their roles.

The industry is experiencing a similar trend. According to Google, 75% of its new code is now AI-generated, highlighting a significant shift in software development approaches. Meta has mandated that 65% of engineers in its creation organization use AI to generate over 75% of their committed code by the first half of 2026. Amazon has integrated Claude Code into its workforce through AWS Bedrock, further exemplifying the growing reliance on AI in programming tasks.

Cherny envisions a future where engineers evolve into "builders," focusing on conceptualizing what needs to be created and guiding AI to realize those visions, rather than engaging in manual coding. This represents a structural shift in the space, and the terminology used to describe it will significantly shape perceptions around these changes. As the industry adapts, Cherny's quest for a more fitting term underscores the urgency to define this new era of software engineering.

Whether the eventual label emerges from current suggestions or a completely new idea, it must encapsulate the reality of AI's influence in coding practices. As the conversation develops, it will be interesting to see how the industry responds to this evolving narrative and what implications it holds for the future of software development.